Read DeepSeek Liang Wenfeng's views on quantitative investment in one minute! ! !
High win rate + high frequency + quantitative decision-making
It took me 5 hours to compile this article. It may be helpful for you to do trading after reading it.
Quantified Trends
First of all, one way to predict the future of quantitative investment in China is to look at the current situation in the United States.
There are two trends in asset management in the United States: one is the gradual indexation of mutual funds, and the other is the gradual quantification of hedge funds.
Foreign hedge funds are equivalent to China's private equity. Initially, hedge funds were not quantitative. This table shows the top 10 global hedge funds in terms of asset management scale in 2004. You can see that most of them are not quantitative.
This is the ranking from 2018, where quantitative funds have already occupied the majority; the well-known Bridgewater ranks first, AQR ranks second, and Renaissance ranks fourth. In recent years, quantitative funds have gradually become the mainstream of hedge funds in the U.S., and many people even believe that hedge funds are just quantitative funds. We are a hedge fund, so today we mainly discuss quantitative funds within hedge funds.
Scale
In recent years, quantitative funds have gradually become the mainstream of hedge funds in the U.S., and many people even believe that hedge funds are just quantitative funds. We are a hedge fund, so today we mainly discuss quantitative funds within hedge funds.
From the experience in the U.S., the management scale of quantitative private equity can be very large. The world's largest hedge fund, Bridgewater, manages about 1 trillion RMB, while large domestic quantitative companies manage between 10 to 20 billion; we may still have several times the growth potential.
Can private equity in China really manage 1 trillion? It should be possible. In the future, China's economy will be comparable to that of the U.S., and the largest domestic teams should be able to manage 200-300 billion. If the stock market expands and the derivatives market develops, they could manage 400-500 billion. Adding in overseas markets, it could reach 1 trillion.
With so many quantitative companies abroad, what are they all doing? Are they all into high frequency? Obviously not. High frequency accommodates very little money; it is not the mainstream of asset management. The answer is that all strategies are being implemented, from macro hedging to stock fundamentals, to stock volume-price, to commodities, to bonds, with the main battleground being stocks and bonds.
The world's largest hedge fund, Bridgewater, focuses on macro quantification, while the second largest, AQR, focuses on stock fundamentals. You will find that the lower the frequency of the strategy, the larger the capacity. All strategies originally executed by humans are now being quantified.
Currently, domestic hedge funds primarily focus on volume-price strategies; overall, we are lagging behind the U.S. From the experience in the U.S., we still have a lot of room for development in terms of strategy types.
So how do we distinguish between quantitative and non-quantitative? Based on China's national conditions, we define quantitative investing. Some say quantitative investing is about programmatic ordering, which is incorrect, as many quantitative companies use manual ordering, while many traditional public funds use programmatic ordering and have mature VWAP systems.
DeepSeek: VWAP (Volume-Weighted Average Price) is a widely used benchmark price in financial markets, reflecting the average trading price of a security over a certain period, weighted by trading volume. VWAP systems are typically used for algorithmic trading, execution strategies, and trading performance evaluation, helping investors complete large transactions at levels close to the market average price, reducing market impact.
Some say that research is conducted using quantitative methods? This is also incorrect, as modern investment research often requires quantitative methods, and this definition lacks distinction; everyone can claim to be quantitative.
Some say that subjective investing requires deep analysis of individual stocks, while quant investing does not need to focus on individual stocks. This is also incorrect; at least we analyze individual stocks quite thoroughly, and our American peers also analyze individual stocks in great detail.
So what is the real difference? The answer lies in whether you are using quantitative methods for decision-making or relying on human decisions. Therefore, the difference is not in trading or research methods, but in decision-making processes.
Quantitative companies also have many traders and researchers, but you will find that quantitative companies do not have fund managers; the fund manager is a bunch of servers. When humans make investment decisions, it is an art and relies on intuition. When programs make decisions, it is a science, and it has an optimal solution.
Some people ask, will quantitative investing still need humans in the future? Of course, it will; a large number of programmers and researchers will be needed.
Next, let's look at what everyone is doing in China's quantitative investment.
Currently, we estimate that the quantitative funds invested in the Chinese market are between 250 billion and 500 billion. More than half of this is invested in stock strategies, followed by commodity CTAs, with the remainder being very small.
From historical returns, stock returns are slightly better than commodity CTAs. Today, we focus on discussing stock strategies. This table was estimated in collaboration with peers; it may not be precise, but the general outline is similar. If you want to invest in quantitative funds, you can refer to this table when looking for an investment advisor.
For stock strategies, we traditionally classify them into four types, with the most important being the first type, the daytime volume-price model. The multi-factor and alpha approaches that everyone often hears about actually refer to this daytime volume-price model, which has a scale of about 200 billion.
The second most important is the intraday reversal model, commonly known as stock T0, which has several hundred billion.
There are also two other types: fundamental models and event-driven models, which are not the focus right now. This is data from private equity, and there are also about 120 billion in public funds doing fundamental quantification; today we are only discussing private equity.
All four of these models are effective. Traditionally, all models are multi-factor models, aiming to obtain excess returns through stock selection and timing.
Before 2017, multi-factor models were universal; we used to hope to replicate the worldquant model by hiring many people to mine factors. Peers were competing over whose factors were more effective. Now, it has become very difficult to extract effective factors.
After 2017, the industry underwent changes, and the traditional multi-factor framework was gradually replaced by artificial intelligence.
After 2019, it has gradually been replaced by newer integrated frameworks.
As a private equity firm, investors have very high expectations of us; if we underperform the index by less than 25% in a year, investors are dissatisfied.
Competition among private equity is intense. We receive performance data from peers every week, comparing who outperformed whom; if we fall behind, clients will immediately call, so we feel a lot of pressure. I believe everyone in the industry feels this pressure.
It is precisely this pressure that forces us to continuously improve our investment capabilities, working overtime to revise strategies, because any laziness will lead to falling behind. Of course, we charge our clients high fees, far exceeding those of public funds, so this performance and pressure are also fair.
We are often asked: who is quant investing really making money from? The answer is quite simple: quant is making money from the original profits earned by human investors.
Human investors are divided into two schools: one is technical analysis, and the other is fundamental analysis. To be more specific, quantitative methods are currently making money that was originally earned by the technical analysis school.
Who can tell me, who made money from the technical analysis school? It is now much harder for the technical analysis school to make money than before, because there are 200-300 billion in programs doing the same thing every day, significantly increasing market efficiency.
In a few years, it will be more difficult for humans because programs are continuously advancing. It is now 2019, and technically, programs have far surpassed the average human expert.
Who can tell me, who made money from the technical analysis school? It is now much harder for the technical analysis school to make money than before, because there are 200-300 billion in programs doing the same thing every day, significantly increasing market efficiency.
In a few years, it will be more difficult for humans because programs are continuously advancing. It is now 2019, and technically, programs have far surpassed the average human expert.
The entire industry of quantitative private equity is progressing roughly in line with Moore's Law, doubling its investment capability every 18 months. However, in recent years, the average return of quantitative investments has remained relatively unchanged due to the continuous improvement in market efficiency. This is logical because if investment capability doubles while market efficiency remains the same, returns should also double. Therefore, market efficiency has improved.
The increase in market efficiency is evidenced by the fact that it has become very difficult for human experts to make money. Another evidence is that quant strategies that were effective two years ago are also gradually becoming ineffective.
The investment capability in quant has a lot of room for improvement. Therefore, we expect that in the coming years, the efficiency of China's stock market will further increase. This is an unstoppable historical trend.
We are often asked a question: if the market becomes very efficient in the future, will everyone stop making money?
From the situation in the U.S., the market will not be 100% efficient. Because if the market were 100% efficient, hedge funds would disappear; who would maintain liquidity and pricing?
The market will reach a balance close to full efficiency, allowing hedge funds to cover their operating costs and the costs of client funds and risks. Globally, hedge funds are not a highly profitable industry when compared to the primary market and real estate.
The historical stage we are in is roughly here; we are still far from a fully efficient market. At least for the next few years, we do not need to consider this issue.
Finally, we make two predictions: one is short-term, and the other is long-term. If these two predictions hold true, the returns of quantitative investments can continue for several more years.
The short-term prediction is for the next one to two years. The improvement in the industry over the next one to two years should come from a combination of multiple strategies.
Combining multiple strategies is not simply about diversification. Diversification would mean having 400 million in funds, with 100 million allocated to model A, 100 million to model B, 100 million to model C, and 100 million to model D. The drawback of this approach is that the return is the average of the four models.
What we mean by combining multiple strategies is layering: 400 million will be used for model A while at the same time, the same 400 million will also be used for models B, C, and D, ultimately synthesizing into one large, comprehensive strategy that does not belong to any traditional category. The return is an integration of the four models.
Last year, daytime alpha combined with intraday T0 performed very well, but it has already lagged, and now more strategies are needed, using more advanced methods for combination. This sounds very reasonable, but it is difficult to implement; the challenge lies not in the strategy or technology itself, but in the business logic of the private equity firm.
Because each model requires a team, where originally one team could manage tens of billions, now multiple teams are needed to manage the same amount, significantly increasing costs, while company revenue does not increase proportionally. However, based on our observations, this trend is already occurring because if you don't do it, others will. Recently, the best-performing private equity firms have all adopted multi-strategy approaches.
We expect this process to accelerate because as market efficiency improves and returns decline, achieving good returns with a single strategy has become very difficult. In the future, strategies as a whole will be extremely complex, with a large workload and high thresholds. Quantitative companies that lack the ability to organize multiple teams will find it difficult to survive.
Quantitative investment will concentrate on leading firms, allowing them to have sufficient resources to execute these more complex strategies. We believe there is still significant space for combining multiple strategies; based on our own pace, we won't be able to finish this in the next one to two years. If this prediction holds true, quantitative private equity can still achieve relatively good returns in the next one to two years.
Long-term forecasts predict the future over the next 3 to 5 years. One day, the volatility in the technical aspects will diminish, as technological advancements reach a bottleneck. In the future, quantitative investment will surely split the profits originally earned by fundamental investors.
In terms of fundamentals, market efficiency is still relatively poor, leaving a lot of room for improvement. Quantitative approaches to fundamentals are technically completely feasible. Some say that fundamentals vary by company and cannot be quantified, which is incorrect.
First, if the U.S. can quantify, why can't China? Secondly, if technical aspects can be quantified, why can't fundamentals?
Around 2015, fundamental quantification was popular among private equity, when market efficiency was not as high as it is now, allowing traditional multi-factor frameworks to be profitable. However, starting in 2017, returns gradually declined, and private equity teams focusing on fundamental quantification lost their competitive edge and were gradually eliminated, while public funds are still doing it.
Private equity needs to elevate fundamental quantification to a higher level. This mission will not be accomplished by the older generation, but by newer, more capable individuals using more complex and refined methods to achieve this.
In our current products, we have already integrated fundamental quantification models with very good results, but we are still using traditional methods. To advance further, we need to refine our approach, which is significantly more costly than technical aspects. To reach a level like AQR, we conservatively estimate that the team's costs exceed 1 billion RMB annually, so we can only take it step by step.
If quantitative private equity can manage 100 billion in the future, this cost is acceptable, and there are no issues with the business model. Fundamental quantification still has a long way to go; it needs to reach the current level of technical quantification, which may take several cycles of Moore's Law. But this day will surely come within our lifetime.
The final question is, if hedge funds are making money from both technical and fundamental aspects, what about ordinary people?
Hedge funds only earn money from volatility, liquidity, and pricing, not from beta. The largest hedge fund in the U.S., Bridgewater, has an asset management scale of 1 trillion RMB, while the largest mutual fund, BlackRock, has an asset management scale of 45 trillion. In front of mutual funds, hedge funds are just small players. When the market is efficient, you can simply buy the index, which is true value investing; the wealth still lies primarily in the hands of the common people.
The above is Liang Wenfeng's understanding of quant from DeepSeek six years ago; China's quantitative market has made significant progress.
$BTC $BNB
